Planarization of semiconductor substrates is becoming more important as the number of layers used to form a semiconductor device increases. Nonplanar semiconductor substrates have many problems including difficulty in patterning a photoresist layer, formation of a void within a film during the film deposition, and incomplete removal of a layer during an etch process leaving residual portions of the layer, which are sometimes called "stringers." A number of planarization processes have been developed and include chemical-mechanical polishing.
Chemical-mechanical polishing may use a polyurethane polishing pad. In chemical-mechanical polishing, a substrate is pressed against a polishing pad that is wetted with a polishing slurry. FIG. 1 includes a cross-sectional view of a portion of a typical polishing pad surface 10. The pad surface 10 consists of many peaks or "asperities" that protrude from the polishing pad surface 14. An asperity 11 is characterized by a height 12 and a radius 13. The height 12 is the distance between the tip of a peak and the polishing pad surface 14. The radius 13 is the radius of curvature at the peak. Heights and radii of the asperities usually vary randomly. A typical commercially-available polishing pad has a standard deviation of heights of about 30 microns and an average radius of about 35 microns.
The random variation of the surface of a polishing pad is believed to be caused in part by the manner in which the polishing pads are formed. Typically, a cylinder of polyurethane is formed and is cut into slices to form a plurality of polishing pads. A cutting blade may snag on polyurethane filaments within the polishing pad causing the random variations of the topography of the polishing pad. This random variation in heights and radii of the asperities in typical polishing pads creates difficulties in substrate polishing.
FIG. 2. includes a cross sectional view of a portion of a semiconductor substrate 20 having a patterned layer 21 and a layer 22 overlying the patterned layer 21. The substrate 20 may include an insulating layer, a conductor, or the like. Each of the patterned layer 21 and layer 22 may be an insulating layer or a conductive layer, but patterned layer 21 and layer 22 are typically dissimilar materials. The layer 22 has a thickness at least as thick as the patterned layer 2.1, so that the openings within the patterned layer 21 are filled. In theory, the layer 22 is to be polished such that layer 22 only remains within the openings, and the combination of the patterned layer 21 and layer 22 forms a fiat surface after polishing.
FIG. 3 shows how varying asperities may cause problems when layer 22 is polished with a polishing pad that includes asperities 31 and 32. The asperities 31 and 32 are located adjacent to each other on the same surface of a polishing pad. The patterned layer 21 has openings that are about the same dimensions. The asperity 31 has a radius that is larger than the openings. As a result, the asperity 31 does not significantly dig into the member 221, which is that portion of layer 22 lying within the left-hand opening. Unlike asperity 31, asperity 32 has a radius that is smaller than the openings. Unfortunately, asperity 32 can dig into member 222, which is that portion of layer 22 lying within the right-hand opening, and causes the well-known problem called "dishing" which is a condition where the thickness of member 222 is thinner near the center of the opening compared to the edge of the opening. Dishing is undesired.
Beyond the problem just described, the difference in heights and radii also cause variations in contact area. In other words, the local polishing rate for asperity 31 is different from the local polishing rate for asperity 32. Because of height and radius variations of the asperities, it is difficult to control the contact properties, such as contact pressure and contact area, between the asperities and portions of the patterned layer 21 or the layer 22. Therefore, random variation of asperity sizes may reduce the ability to selectively remove layer 22 uniformly across the substrate.
A perfectly fiat sheet of polishing pad is not expected to solve the prior art problems. Even at relatively low platen and/or substrate rotational speeds, a thin film of slurry may form between the substrate and the polishing pad and cause hydroplaning. With hydroplaning, any polishing that may occur is primarily only chemical and not mechanical. This is not desired because the removal rate of the layer 22 declines. Also, the polishing is likely to become less preferential. In other words, the polishing may remove the higher portions of layer 22 at a rate that is closer to the rate that the lower portions of layer 22 are removed. Therefore, hydroplaning may cause a lower selectivity between the high portions and low portions of a layer to be polished which is undesired.